Ad

MTa

b

c

d

e

a

ARRAA

KBTEGC

1

aidifpoo

L[ost

S1

(

0d

Journal of Membrane Science 382 (2011) 186– 193

Contents lists available at ScienceDirect

Journal of Membrane Science

j ourna l ho me pag e: www.elsev ier .com/ locate /memsci

case study of the effect of grain size on the oxygen permeation flux of BSCFisk-shaped membrane fabricated by thermoplastic processing

ehdi Salehia,b,∗, Frank Clemensa, Ewald M. Pfaff c, Stefan Diethelmd, Colin Leache,homas Graulea, Bernard Grobétyb

Laboratory for High Performance Ceramics, Empa, Swiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse 129, CH-8600 Duebendorf, SwitzerlandThe Fribourg Center for Nanomaterials (FriMat) and Department of Geosciences, University of Fribourg, Pérolles Ch. du Musée 6, CH-1700 Fribourg, SwitzerlandInstitute for Materials Applications in Mechanical Engineering, RWTH Aachen University, Nizzaallee 32, 52072 Aachen, GermanyLaboratory for Industrial Energy Systems (LENI), STI, Swiss Federal Institute of Technology, 1015 Lausanne, SwitzerlandSchool of Materials, The University of Manchester, Manchester M13 9PL, United Kingdom

r t i c l e i n f o

rticle history:eceived 31 May 2011eceived in revised form 19 July 2011ccepted 4 August 2011vailable online 11 August 2011

eywords:

a b s t r a c t

Oxygen permeability measurements of Ba0.5Sr0.5Co0.8Fe0.2O3−ı (BSCF) disk shaped membranes fabricatedby thermoplastic processing and sintered at different temperatures (1000–1100 ◦C), showed no influ-ence of the grain size on the oxygen permeation fluxes. To further investigations, Electron BackscatteredDiffraction (EBSD) and Conductive mode (CM) microscopy methods were used for texture analysis andobservation of the local electrical behavior in the BSCF membranes, respectively. EBSD results revealedthat the grain size of the membranes increased with increasing the sintering temperature from an aver-

◦ ◦

SCF membranehermoplastic processingBSD methodrain sizeM analysis

age of 3.32 �m at 1000 C to 18.25 �m at 1100 C. Also, it was seen that there was no textural differencebetween the different samples. CM analysis demonstrated that the electronic conductivity of the grainsand grain boundaries was similar in the membrane sintered at 1000 ◦C. Finally, the stability of the mem-brane under the operation conditions was tested, and it was found that the permeation flux was nearlyconstant at 900 ◦C after an operation time of more than 50 h, whereas oxygen permeation flux declined

at 82

after a relative short time

. Introduction

Perovskite-based, mixed ionic–electronic conductors (MIEC)re of great interest because of their 100% oxygen permselectiv-ty at elevated temperatures. These materials are widely underevelopment as ceramic membranes for oxygen separation and

n catalytic partial oxidation reactors, as electrodes for solid oxideuel cells and for high temperature electrolysis [1–4]. Oxygenermeation in MIEC membranes is driven solely by a difference ofxygen partial pressure (oxygen potential gradient) on either sidef the membranes.

Since Teraoka reported high oxygen permeation flux ina1−xSrxCo1−yFeyO3−ı (LSCF) ceramic membranes in the late 1980s5,6], a large number of studies have been carried out to optimize

xygen permeation in this system, with SrCo0.8Fe0.2O3−ı (SCF)howing significant promise. However, later studies revealedhat the SCF phase was not stable below 800 ◦C at oxygen partial

∗ Corresponding author at: Laboratory for High Performance Ceramics, Empa,wiss Federal Laboratories for Materials Science and Technology, Ueberlandstrasse29, CH-8600 Duebendorf, Switzerland. Tel.: +41 58 765 4952; fax: +41 44 823 4150.

E-mail addresses: Mehdi.Salehi@empa.ch, chemehdi3333@gmail.comM. Salehi), Frank.Clemens@empa.ch (F. Clemens).

376-7388/$ – see front matter © 2011 Elsevier B.V. All rights reserved.oi:10.1016/j.memsci.2011.08.007

5 ◦C.© 2011 Elsevier B.V. All rights reserved.

pressures lower than about 0.1 atm due to oxygen vacancy ordering[5,7–10].

It was later found that the structural and chemical stability of theSCF membranes were greatly improved by partial substitution of Srwith Ba, retaining a high oxygen permeation flux for compositionsaround Ba0.5Sr0.5Co0.8Fe0.2O3−ı (BSCF). Recently, BSCF has receivedincreasing attention for use in ceramic oxygen membranes [11–15].

It has been reported that the oxygen permeation flux of the MIECmembranes is sensitive to microstructural features, such as grainsize and grain boundary structure [16–26]. The microstructuredepends both on processing history, and especially the sinteringprocedure, even for similar components. The preparation of MIECmembranes can be considered in terms of three steps: (1) powdersynthesis, (2) shaping, and (3) sintering. The influence of each ofthese steps on the microstructure and oxygen permeation flux ofthe MIEC membranes has been investigated over the last decade,with several different behaviors reported [13,16,17,19–21,24].

In the case of BSCF, previous studies have resulted in conflict-ing data, especially with regard to the influence of grain size onpermeation [13,14,18,22–26]. The effect of the microstructural on

the oxygen permeation flux cannot be generalized and the contra-dictions could be attributed to microstructural variations causedby differences in powder synthesis routes and sintering conditionsused by different groups.

M. Salehi et al. / Journal of Membran

obgtubmtc

2

2

(cawba

rCAa

Fig. 1. Schematic diagram of the BSCF membrane fabrication.

In this study, we have examined the effect of grain sizen the oxygen permeation flux of Ba0.5Sr0.5Co0.8Fe0.2O3−ı mem-ranes, fabricated by warm pressing. Three samples with differingrain sizes were prepared by sintering at different tempera-ures. The associated microstructural changes were characterizedsing scanning electron microscopy (SEM) combined with electronackscatter diffraction (EBSD) for texture analysis and conductiveode (CM) microscopy to observe the local electrical behavior in

he BSCF membrane. Oxygen permeation experiments were alsoarried out on the membranes sintered at each temperature.

. Experimental

.1. Sample preparation

Commercially available Ba0.5Sr0.5Co0.8Fe0.2O3−ı (BSCF) powderTreibacher, Austria), with a mean particle size of 3 �m and a spe-ific surface area of 1.55 m2/g was used. Mean particle size (dBET)lso was calculated from the BET according to dBET = 6/[�thSBET],here �th is the theoretical density of the BSCF powder measured

y He-pycnometer and SBET is the measured surface area by BETnd a value of 0.70 �m was determined.

Fig. 1 shows the thermoplastic processing route used for fab-

icating the disk shaped membranes. Polystyrene (type 648, Dowompany, Switzerland) and polyethylene glycol (PEG 20000) (FlukaG, Switzerland) were added as binders, with additions of steariccid as a surfactant (Fluka AG, Switzerland) in the preparation of the

e Science 382 (2011) 186– 193 187

feedstock, forming a ceramic–polymer composite with 48% volumefraction of polymer. This is in agreement with literature reports thatthe usual powder content attained is around 52 vol.% for ceramicfeedstock with binders based on thermoplastic polymers [27]. Thefeedstock was prepared using a high shear mixer (HAAKE PolyLabMixer, Rheomix 600, Thermo Scientific, Germany). The mixing wascarried out using a two step sequence: in the first step, the mixingwas performed at 10 rpm and 170 ◦C for 30 min and subsequentlyat 10 rpm and 150 ◦C for 150 min until the mixing torque reacheda constant value.

The feedstocks were loaded into a steel die with heating man-tle and uniaxially pressed at 30 kN and 165 ◦C for 30 min to formdisk-shaped membranes, approximately 28 mm in diameter. Theelimination of the binder was achieved by thermal debinding,employing a slow heating rate up to 600 ◦C in air. To provide sam-ples with different grain size, sintering was carried out at 1000 ◦C,1050 ◦C and 1100 ◦C for 2 h, using a heating rate of 1 ◦C/min. Thedensities of the sintered samples were measured by the Archimedesmethod. The porosity of the sintered disks was determined usingimage analysis (Digital Micrograph 3.10.0, Gatan Inc.). Additionallythe porosity was calculated by the quotient of the Archimedes den-sity and the skeleton density (He-pycnometer) and subtracting theresults from one.

2.2. Basic characterization

2.2.1. Electron Backscattered Diffraction (EBSD)EBSD is widely used in quantitative metallography; particularly

for subgrain imaging and texture determination [28–30]. Its usein ceramics research is increasing but is limited by difficulties inproducing the necessary strain-free surface to allow channeling ofelectrons to take place. A further benefit is that orientation contrastallows grains to be differentiated easily without sample etching.

Kikuchi-type diffraction patterns, generated by backscatteredelectrons [28–30], are recorded for each point of analysis. Thecrystallographic orientation is determined after indexing of thecorresponding diffraction pattern. The structure factors of thereflectors are calculated with Electron Microscopy Image Simula-tion (EMS) software developed by Stadelmann [31]. For the EBSDsample preparation, the sintered samples were ground and pol-ished to achieve a smooth strain-free surface. All EBSD data werecollected at the University of Fribourg using a Philips® FEI XL30SFEG Sirion SEM equipped with an EDAX® (TSL) OIM 3.5 softwarepackage. The best results were obtained with an accelerating volt-age of 15 kV and a probe current of 20 nA. The sample tilt was 70◦

and working distance of 15 mm. To reduce the measuring time, theresolution, i.e. the step size and analyzed area were adjusted tomatch the average grain size. For the fine grained material sinteredat 1000 ◦C an area of 120 �m × 120 �m was mapped with a step sizeof 0.25 �m/step, whereas for coarse grained membrane sintered at1100 ◦C an area of 570 �m × 264 �m was mapped at a resolution of1.25 �m/step.

Orientation distribution functions were calculated for the(1 0 0), (1 0 1) and (1 1 1) axes. To estimate the strength of the LatticePreferred Orientation (LPO), the texture index J, which is the meansquare value of the orientation distribution function, was deter-mined [32]. A purely random LPO gives J = 1.0, with J increasing withthe degree of preferred orientation. For a single crystal, J tends toinfinity (in practice it is limited to about 270 due to the truncationin the spherical harmonics calculations).

The grain misorientation distribution has been determined fromthe ODF (Orientation Distribution Function) and compared to ran-

dom distributions [33]. The grain boundaries (GB’s) satisfying thecoincidence site lattice (CSL) model for the multiplicity index val-ues up to 49 were automatically detected for cubic materials, usingthe list integrated in the software [34]. The multiplicity index is

188 M. Salehi et al. / Journal of Membrane Science 382 (2011) 186– 193

nt dev

daTtstocapm

2

M[etsaaotalaoAgrsflmr

2

i[eaiaa

at

tering temperature.Fig. 4 shows Secondary Electron (SE) images of the mem-

brane samples sintered at each temperature, with superimposed

electron beam

to ampli fierto ea rth

prob e contact

Fig. 2. Schematic picture of setup of the measureme

efined as the reciprocal ratio between the total number of sitesnd the number of coincident sites of two interpenetrating lattices.he latter form the coincident site lattice. The smaller the valuehe higher the coincidence between both lattices. The boundarieseparating neighboring grains, for which the interpenetrating lat-ices form a CSL with low index, are often parallel to planesf the latter. Special GB’s corresponding to given values wereounted and expressed as a fraction of the total GB’s. The numbernd type of such special boundaries determined in the present sam-les have been compared with the values calculated for a randomicrostructure.

.2.2. Permeation measurementsThe set-up used for permeation experiments is shown in Fig. 2.

easurements were carried out as has been described previously35]. The sintered disc membranes, approximately 23 mm in diam-ter and 0.6 mm in thickness, were clamped between two aluminaubes and sealed using gold rings and paste. The surface of theamples was ground before permeation measurement. Syntheticir (350 Nml/min) was pumped into the lower compartment, andrgon (99.999%) (100 Nml/min) was used as sweep gas in the upperne. Oxygen was separated from air by solid state diffusion throughhe dense membrane due to the oxygen partial pressure gradientnd released on the argon side. The outlet gas composition was ana-yzed with a gas chromatograph (GC) (Varian Inc.) equipped with

molecular sieve, 5 A capillary columns and a TCD detector. Theutlet flux was measured using a digital flow-meter (Varian Inc.).ny leakage could be detected from the presence of N2 in the outletas. The sample was first heated up to 1000 ◦C to soften the goldings and ensure a good seal. The temperature was then decreasedtep-wise, with permeation measurements made every 25 ◦C. Theux was measured after equilibration of the sample. Flux measure-ents were repeated up to 7 times for each condition to check the

eproducibility.

.2.3. Conductive mode (CM) microscopyThe conductive mode (CM) of the SEM uses currents flowing

n the sample under beam irradiation as the image forming signal36] and is used in the characterization of semiconductors and somelectroceramics [37–41]. In use, two current collecting electrodesre positioned on the sample surface to either side of the area ofnterest, permitting currents that are generated within the samples a result of primary electron beam bombardment to be collected

nd amplified to form the image (Fig. 3).

CM microscopy has been used to observe local resistivity vari-tions in electroceramics using resistive contrast (RC) imagingechniques [37,40,41]. In this mode of operation a pair of electrodes

ice for oxygen permeation flux of membranes [35].

is deposited on either side of the area of interest, making electricalcontact. One electrode is connected directly to earth and the otherconnected to an amplifier so that the absorbed beam current flow-ing through it can be used to control the brightness of the RC imageon the screen. As the electron beam scans the area between the elec-trodes, injecting current, the sample acts as a current divider andvaries the proportion of current flowing to each electrode (Fig. 3).In a uniformly resistive specimen the current flowing to the ampli-fier and hence the screen brightness varies smoothly. However, alocal change in specimen resistivity changes the rate of variationof current reaching the amplifier, and is seen as a perturbation inthe brightness gradient in the RC image. The resolution of the tech-nique is limited by the size of the electron excitation volume in thespecimen, and is therefore dependent on beam energy and spec-imen density, and in the case of our material and experimentalconditions is approximately 1.1 �m. To observe RC in our sam-ple, pairs of 100 �m square, gold electrodes, separated by 50 �mwere applied to a polished face of the BSCF sample through aphotolithography-applied mask. The samples were mounted onan electrically insulating sample holder in a JEOL 6300 SEM, andconnected to the amplifier via tungsten probes controlled by amicromanipulator. The CM signal collected under primary beamirradiation (10 keV, 5 nA) was amplified for imaging and linescananalysis using a commercial EBIC amplifier (Gatan EbiQuant).

3. Results

Table 1 gives the sintered density, average grain size and poros-ity of the BSCF membranes sintered at different temperatures. Thesintered densities and grain size both increase with increasing sin-

BSCF specimeninsulating sub strate

Fig. 3. CM configuration.

M. Salehi et al. / Journal of Membrane Science 382 (2011) 186– 193 189

Table 1Effect of the sintering temperature on the densification of the membrane.

Sinteringtemperature(◦C)

Sintereddensityb

(g/cm3)

Sintereddensityb (%theoretical)

Porositya,b

(%)Average grainsizec (�m)

1000 5.29 ± 0.02 96.27 ± 0.36 3.45 ± 0.46 3.321050 5.32 ± 0.02 96.63 ± 0.15 3.25 ± 0.28 6.861100 5.38 ± 0.01 97.91 ± 0.25 1.82 ± 0.42 18.25

a Porosity was determined from measured density (Archimedes method) andskeleton density (pycnometer).

b

t

Eettadaaimlpts

Fm

The average and deviation values were reported for three samples at each sin-ering temperature.

c Average grain size was determined from the EBSD analysis.

BSD orientation maps, in which each grain’s crystallographic ori-ntation relative to the image normal is given by the color inhe unit triangle. (For interpretation of the references to color inhis sentence, the reader is referred to the web version of therticle.) Grain size distribution of samples was determined by10 ≈ 2 �m and d90 ≈ 7 �m after sintering at 1000 ◦C/2 h, d10 ≈ 6 �mnd d90 ≈ 17 �m after sintering at 1050 ◦C/2 h, as well as d10 ≈ 12nd d90 ≈ 40 for the sample sintered at 1100 ◦C/2 h. The black areasn the SE image, which are not indexed in the EBSD orientation

aps, correspond to pores, which are 1–10 �m in diameter and areocated both within the grains and at the grain boundaries. The pro-

ortion of intra-granular pores increases with increasing sinteringemperature. The overall porosity (determined by image analy-is) was about 11.8%, 8.8% and 6.1% for the membranes sintered

ig. 4. SE images with superimposed grain orientation maps of cross-sections of theembranes sintered at: (a) 1000 ◦C, (b) 1050 ◦C, and (c) 1100 ◦C.

Fig. 5. Grain size distribution of the membranes sintered at different temperatures.

at 1000 ◦C, 1050 ◦C and 1100 ◦C, respectively. The porosity deter-mined by image analysis showed a higher value than the porosityobtained by measured density (Table 1). The difference could bedue to the threshold values set during the image analysis. Fig. 5shows the grain size distributions within each sample, determinedfrom the EBSD imaging.

In all three samples the grains are randomly oriented, with val-ues of the texture index, J, of 1.17 (1000 ◦C), 1.85 (1050 ◦C) and 1.21(1100 ◦C) for the (1 0 0) poles (Fig. 6). The grain boundary misorien-tation angle distributions follow the expected curves for a randomtexture in a cubic material (Fig. 7).

The fractions of special CSL boundaries (Fig. 8) are close to thedistribution calculated for a random texture [33], i.e. they con-firm the lattice preferred orientations (LPO) results, but are slightlyhigher than expected, although this may be due to parameters usedin applying the Brandon criterion [42], which was originally derivedto accommodate the effects of tilting at grain boundaries due to dis-location arrays. These microstructural data suggest that the effectof changing the sintering temperature has merely been to adjustthe grain size. There is no evidence for changes in texture, or grainboundary structure distribution that might affect permeation rate.The lack of a discernable texture also means that a random distri-bution of crystal surfaces will be presented at the surface of themembrane.

Fig. 9 depicts the oxygen permeation flux of the BSCF mem-branes sintered at 1000, 1050 and 1100 ◦C for 2 h as a function ofthe temperature. The error bars correspond to the standard errorcalculated from the standard deviation of the experimental data.The oxygen permeation fluxes through the membranes are iden-tical within the error margins of the measurement. Our values forthe oxygen permeation fluxes at 950 ◦C and 850 ◦C are comparedwith those of other Ba0.5Sr0.5Co0.8Fe0.2O3−ı (BSCF) disk membranesreported in the literatures (Table 2). The apparent activation energyfor permeation in the temperature range 800–1000 ◦C was foundto be ∼40 kJ/mol for all of the membranes used in this study. Thisvalue is consistent with other reported values for bulk diffusion inBSCF membranes, which reported the apparent activation energybetween 25 and 45 kJ/mol in the temperature range of 800–950 ◦C[11,26,43,44].

Fig. 10a shows a secondary electron image of the polished faceof the BSCF membranes sintered at 1000 ◦C and prepared for CMmicroscopy. The surface electrodes and electrical contact probes

are visible in the image. Fig. 10b is a CM image of the same area.The image brightness is given by the proportion of absorbed beamcurrent flowing to the top electrode. The bright contrast under thetop electrode and the dark contrast under the lower electrode is

190 M. Salehi et al. / Journal of Membrane Science 382 (2011) 186– 193

Fig. 6. Pole figures for the (0 0 1), (1 0 1) and (1 1 1) axes of the 1000 ◦C sample (max. value is very low). The results for the other two samples are similar (RD: rolling directionand TD: transverse direction).

Fig. 7. Grain boundary misorientation angle distributions for (a) 1000 ◦C, (b) 1050 ◦C, and (c) 1100 ◦C. The blue curve is the distribution calculated for a cubic symmetrysample with random texture. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

F d to Ci to the

ctfwvitibe

ig. 8. Grain boundary map for 1050 ◦C sample. The colored boundaries corresponnterpretation of the references to color in this figure legend, the reader is referred

aused by an electron beam induced current (EBIC) arising fromhe partial Schottky nature of the gold contacts on the sample sur-ace. Between the electrodes, the image has a uniform grey gradientith no microstructural detail. The graph in Fig. 10c shows that the

ariation in CM current with beam position between the electrodess linear and since any sudden local change in resistivity would dis-

ort the uniform variation of CM current with beam position. Thisndicates that the sample resistivity is uniform on the scale of theeam excitation volume within this area. The injection of beamlectrons changed the resistance between the electrodes by less

SL boundaries with = 3 (dark blue), 5 (light blue), 7 (light green), ≥9 (red). (For web version of the article.)

than 1% and so the sample free-carrier concentration was not sig-nificantly altered by the measurement. It is therefore concludedthat the electronic conductivity of the grains and grain boundariesin this region are similar.

It is important for the membranes to remain stable underthe operating conditions and to show no degradation in oxygen

permeation. Fig. 11 shows the time dependence of the oxygenpermeation flux for the BSCF membrane sintered at 1000 ◦C/2 h. Itwas found that the permeation flux was nearly constant at 900 ◦Cafter an operation time of more than 50 h, whereas oxygen per-

M. Salehi et al. / Journal of Membrane Science 382 (2011) 186– 193 191

Fig. 9. Temperature dependence of oxygen permeation flux through the differ-ent BSCF membranes sintered at 1000 ◦C, 1050 ◦C and 1100 ◦C for 2 h, air flowra

mTttwtt

Table 2Comparison of the oxygen permeation fluxes through the Ba0.5Sr0.5Co0.8Fe0.2O3−ı

(BSCF) disk-type membranes.

Temperature(◦C)

Thickness(mm)

O2 flux(ml/cm2 min)

References

950 1.50 1.60 [1]950 1.80 1.40 [11]850 1–2 2.01 [12]950 0.60 2.43 Present worka

et al. [46]. However, these samples had very low densities, ranging

ate = 350 Nml/min and sweep flow rate 100 Nml/min. The data are average valuesnd the error bars correspond to standard error.

eation flux declined after a relatively short time (20 h) at 825 ◦C.his difference in behavior with temperature may be attributableo a decrease in the concentration of mobile oxygen vacancy inhe lattice owing to changes in the oxidation state of B cations, in

hich, the structure gives a Goldschmidt tolerance factor higher

han one [3,45]. Another reason is the hexagonal-cubic phaseransition that occurs in BSCF at moderate temperatures [11,13]

Fig. 10. (a) Secondary electron image of the polished face of the BSCF, (

850 0.60 1.70 Present worka

a Oxygen permeation fluxes were reported for samples sintered at 1000 ◦C/2 h.

and this phase transformation would be ordered the oxygenvacancies, whose are almost non-oxygen permeable [9].

4. Discussion

The EBSD data show that all three samples have close to ran-dom textures with similar grain boundary structure distributions.The only microstructural parameters that change are grain size andporosity: porosity only slightly. Within the accuracy of our experi-ment, changes in grain size in the range 3–18 �m did not affect theoxygen flux at temperatures in the range 800–1000 ◦C.

Wang et al. [13] reported a positive, albeit weak, correlationbetween flux and grain size, in the range 30–140 �m, which is largerthan used in this study. Similar observations were made by Gao

from 69 to 85% of the theoretical BSCF density. A much strongercorrelation was observed by Arnold et al. [24], but in this case thegrain size (14–17 �m) was controlled by the amount of BN (boron

b) CM image, and (c) variation in CM current with beam position.

192 M. Salehi et al. / Journal of Membran

Fa

ntbtbmaHppIdra

tflaioLaeoaa

tsftc

5

wg1tgcssto

[

[

[

[

[

[

[

[

[

[

ig. 11. Oxygen permeation flux through the BSCF membrane as s function of timet 825 and 900 ◦C for membrane sintered at 1000 ◦C/2 h.

itride) that was added to the powder. Increasing the concentra-ion of BN, which oxidized and melted during sintering, reducedoth the grain size and the oxygen permeability. The authors claim,hat there is no B2O3 or related phase along the grain boundaries,ut they give no indication about where the B2O3 is located in theicrostructure. Tan et al. [18] also observed an increase in perme-

bility with increasing grain size (0.1 and 0.5 �m on the surface).owever, powders with different synthesis methods were used forreparing the membranes and therefore slight variations in com-osition might influence the flux performance of the membranes.

n contrast to all the above examples, Bouwmeester [23] reported aecrease in flux with increasing grain size and Baumann et al. [26]eported an inverse but very weak relationship between grain sizend the oxygen permeation flux for BSCF.

In other related perovskite systems a general explanation abouthe effect of membrane grain size on the oxygen permeationux cannot be found. While in the LaCoO3−ı, La0.3Sr0.7CoO3−ı

nd La0.6Sr0.4Co0.2Fe0.8O3−ı systems oxygen permeation fluxncreases with increasing grain size [47–49], the opposite isbserved for the La0.5Sr0.5FeO3−ı, La0.6Sr0.4Fe0.9Ga0.1O3−ı anda0.1Sr0.9Co0.9Fe0.1O3−ı systems [20,50,51]. Slower ionic transportt the grain boundary is well known as the grain boundary blockingffect, which is mainly attributed to impurities, solute segregation,r second phase at the grain boundary: even in high purity materi-ls [20,52–54]. Thus, the smaller grain size and consequent higherrea of grain boundary lead to lower total conductivities.

In the present work, we have established that for a random tex-ure the oxygen permeation flux in BSCF is not dependent on grainize and/or grain boundary volume, and have found no evidenceor a grain boundary blocking effect in BSCF. The independence ofhe electronic conductivity from intervening grain boundary wasonfirmed by CM measurements.

. Conclusion

BSCF membranes, 0.6 mm in thickness, were fabricated by thearm pressing method using thermoplastic processing. Different

rain sizes were obtained by sintering the samples at 1000 ◦C,050 ◦C and 1100 ◦C for 2 h. A density of the 96% was achieved forhe membrane sintered at 1000 ◦C and 98% for 1100 ◦C. An averagerain size of 3.23 �m for fine grained membrane and 18.25 �m foroarse grained membrane were determined. Permeation flux mea-

urements showed no difference in performance for membranesintered at different sintering temperatures. EBSD revealed thathere is no lattice preferred orientation as indicated by the valuesf the texture index J, for all three membranes. CM method also

[

[

e Science 382 (2011) 186– 193

showed that the electronic conductivity of the grains and grainboundaries in this region are similar. The stability of the mem-branes associated with oxygen permeability was investigated andresults showed that at temperatures lower than 900 ◦C the oxygenpermeation flux declined with time.

Acknowledgements

The authors would like to acknowledge the financial contri-bution provided by Swiss Electric Research and the CompetenceCenter for Energy & Mobility (CCEM) in Switzerland. The authorsalso would like to thank Mr. Christoph Neururer at Université deFribourg for the EBSD measurements.

References

[1] Z. Shao, G. Xiong, H. Dong, W. Yang, L. Lin, Synthesis, oxygen permeationstudy and membrane performance of a Ba0.5Sr0.5Co0.8Fe0.2O3−[delta] oxygen-permeable dense ceramic reactor for partial oxidation of methane to syngas,Separation and Purification Technology 25 (2001) 97–116.

[2] M.D. Mat, X. Liu, Z. Zhu, B. Zhu, Development of cathodes for methanol andethanol fuelled low temperature (300–600 ◦C) solid oxide fuel cells, Interna-tional Journal of Hydrogen Energy 32 (2007) 796–801.

[3] S. Svarcová, K. Wiik, J. Tolchard, H.J.M. Bouwmeester, T. Grande, Structuralinstability of cubic perovskite BaxSr1−xCo1−yFeyO3−[delta], Solid State Ionics 178(2008) 1787–1791.

[4] Z. Chen, R. Ran, Z. Shao, H. Yu, J.C.D.d. Costa, S. Liu, Further performanceimprovement of Ba0.5Sr0.5Co0.8Fe0.2O3−[delta] perovskite membranes for air sep-aration, Ceramics International 35 (2009) 2455–2461.

[5] Y. Teraoka, H.M. Zhang, S. Furukawa, N. Yamazoe, Oxygen permeation throughperovskite-type oxides, Chemistry Letters 14 (1985) 1743–1746.

[6] Y. Teraoka, H.M. Zhang, K. Okamoto, N. Yamazoe, Mixed ionic–electronic con-ductivity of La1−xSrxCo1−yFeyO3−ı perovskite-type oxides, Materials ResearchBulletin 23 (1988) 51–58.

[7] H. Kruidhof, H.J.M. Bouwmeester, R.H.E.v. Doorn, A.J. Burggraaf, Influence oforder–disorder transitions on oxygen permeability through selected nonstoi-chiometric perovskite-type oxides, Solid State Ionics 63–65 (1993) 816–822.

[8] V.V. Kharton, A.A. Yaremchenko, E.N. Naumovich, Research on the electrochem-istry of oxygen ion conductors in the former Soviet Union. II. Perovskite-relatedoxides, Journal of Solid State Electrochemistry 3 (1999) 303–326.

[9] T. Nagai, W. Ito, T. Sakon, Relationship between cation substitution and stabilityof perovskite structure in SrCoO3−[delta]-based mixed conductors, Solid StateIonics 177 (2007) 3433–3444.

10] Z. Rui, J. Ding, Y. Li, Y.S. Lin, SrCo0.8Fe0.2O3−[delta] sorbent for high-temperatureproduction of oxygen-enriched carbon dioxide stream, Fuel 89 (2010)1429–1434.

11] Z. Shao, W. Yang, Y. Cong, H. Dong, J. Tong, G. Xiong, Investigation of the perme-ation behavior and stability of a Ba0.5Sr0.5Co0.8Fe0.2O3−[delta] oxygen membrane,Journal of Membrane Science 172 (2000) 177–188.

12] H. Lu, J. Tong, Y. Cong, W. Yang, Partial oxidation of methane inBa0.5Sr0.5Co0.8Fe0.2O3−[delta] membrane reactor at high pressures, CatalysisToday 104 (2005) 154–159.

13] H. Wang, C. Tablet, A. Feldhoff, J. Caro, Investigation of phase structure,sintering, and permeability of perovskite-type Ba0.5Sr0.5Co0.8Fe0.2O3−[delta]membranes, Journal of Membrane Science 262 (2005) 20–26.

14] J.F. Vente, S. McIntosh, W.G. Haije, H.J.M. Bouwmeester, Properties and perfor-mance of BaxSr1−xCo0.8Fe0.2O3−ı materials for oxygen transport membranes,Journal of Solid State Electrochemistry 10 (2006) 581–588.

15] P. Zeng, Z. Chen, W. Zhou, H. Gu, Z. Shao, S. Liu, Re-evaluation ofBa0.5Sr0.5Co0.8Fe0.2O3−[delta] perovskite as oxygen semi-permeable membrane,Journal of Membrane Science 291 (2007) 148–156.

16] K. Zhang, Y.L. Yang, D. Ponnusamy, A.J. Jacobson, K. Salama, Effect of microstruc-ture on oxygen permeation in SrCo0.8Fe0.2O3−ı , Journal of Materials Science 34(1999) 1367–1372.

17] V.V. Kharton, E.N. Naumovich, A.V. Kovalevsky, A.P. Viskup, F.M. Figueiredo,I.A. Bashmakov, F.M.B. Marques, Mixed electronic and ionic conductivity ofLaCo(M)O3 (M = Ga, Cr, Fe or Ni): IV. Effect of preparation method on oxygentransport in LaCoO3−[delta], Solid State Ionics 138 (2000) 135–148.

18] L. Tan, X. Gu, L. Yang, W. Jin, L. Zhang, N. Xu, Influence of powder syn-thesis methods on microstructure and oxygen permeation performance ofBa0.5Sr0.5Co0.8Fe0.2O3−[delta] perovskite-type membranes, Journal of MembraneScience 212 (2003) 157–165.

19] S. Diethelm, J.V. Herle, J. Sfeir, P. Buffat, Influence of microstructure on oxy-gen transport in perovskite type membranes, British Ceramic Transactions 103(2004) 147–152.

20] S. Diethelm, J. Van herle, J. Sfeir, P. Buffat, Correlation between oxygen transportproperties and microstructure in La0.5Sr0.5FeO3−[delta], Journal of the EuropeanCeramic Society 25 (2005) 2191–2196.

21] A.L. Shaula, R.O. Fuentes, F.M. Figueiredo, V.V. Kharton, F.M.B. Marques,J.R. Frade, Grain size effects on oxygen permeation in submicrometric

bran

[

[

[

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

M. Salehi et al. / Journal of Mem

CaTi0.8Fe0.2O3−[delta] ceramics obtained by mechanical activation, Journal of theEuropean Ceramic Society 25 (2005) 2613–2616.

22] M. Arnold, H. Wang, A. Feldhoff, Influence of CO2 on the oxygenpermeation performance and the microstructure of perovskite-type(Ba0.5Sr0.5)(Co0.8Fe0.2)O3−[delta] membranes, Journal of Membrane Science293 (2007) 44–52.

23] H.J.M. Bouwmeester, Mixed ionic–electronic conducting membranes for oxy-gen separation, in: 9 Jülicher Werkstoffsymposium, Jülich, Germany, 2007.

24] M. Arnold, J. Martynczuk, K. Efimov, H. Wang, A. Feldhoff, Grain boundaries asbarrier for oxygen transport in perovskite-type membranes, Journal of Mem-brane Science 316 (2008) 137–144.

25] M. Zwick, E.M. Pfaff, Microstructure optimization of Ba0.5Sr0.5Co0.8Fe0.2O3–ı

perovskite for OTM reactors, in: 32nd International Conference on AdvancedCeramics and Composites, Daytona Beach, Florida, USA, 2008.

26] S. Baumann, F. Schulze-Küppers, S. Roitsch, M. Betz, M. Zwick, E.M. Pfaff,W.A. Meulenberg, J. Mayer, D. Stöver, Influence of sintering conditions onmicrostructure and oxygen permeation of Ba0.5Sr0.5Co0.8Fe0.2O3−[delta] (BSCF)oxygen transport membranes, Journal of Membrane Science 359 (2010)102–109.

27] F. Clemens, T. Graule, Thin wall ceramic tubes by extrusion ofthermoplastic–ZrO2 compounds Key Engineering Materials, vol. 206–213,Trans Tech Publications, Zurich, 2002, pp. 425–428.

28] D.J. Prior, A.P. Boyle, F. Brenker, M.C. Cheadle, D. Austin, G. Lopez, L. Peruzzo, G.J.Potts, S. Reddy, R. Spiess, N.E. Timms, P. Trimby, J. Wheeler, L. Zetterström, Theapplication of electron backscatter diffraction and orientation contrast imagingin the SEM to textural problems in rocks, American Mineralogist 84 (1999)1741–1759.

29] F.J. Humphreys, Review grain and subgrain characterisation by electronbackscatter diffraction, Journal of Materials Science 36 (2001) 3833–3854.

30] J. Cho, C.M. Wang, H.M. Chan, J.M. Rickman, M.P. Harmer, A study of grain-boundary structure in rare-earth doped aluminas using an EBSD technique,Journal of Materials Science 37 (2002) 59–64.

31] P.A. Stadelmann, EMS – a software package for electron diffraction analysisand HREM image simulation in materials science, Ultramicroscopy 21 (1987)131–145.

32] H.J. Bunge, Texture Analysis in Materials Science, Butterworths, London, 1982.33] P. Vonlanthen, B. Grobety, CSL grain boundary distribution in alumina and

zirconia ceramics, Ceramics International 34 (2008) 1459–1472.34] P.H. Pumphrey, K.M. Bowkett, Axis/angle pair description of coincidence site

lattice grain boundaries, Scripta Metallurgica 5 (1971) 365–369.35] S. Diethelm, J. Sfeir, F. Clemens, J. Van herle, D. Favrat, Planar and

tubular perovskite-type membrane reactors for the partial oxidation ofmethane to syngas, Journal of Solid State Electrochemistry 8 (2004) 611–617.

36] W. Ehrenberg, D.J. Gibbons, Electron Bombardment Induced Conductivity and

its Applications, Academic Press, New York, 1981.

37] C.A. Smith, C.R. Bagnell, E.I. Cole Jr., F.A. DiBianca, D.G. Johnson, W.V.Oxford, R.H. Propst, Resistive contrast imaging: a new SEM mode for fail-ure analysis, IEEE Transactions on Electron Devices ED-33 (1986) 282–285.

[

[

e Science 382 (2011) 186– 193 193

38] J. Palm, H. Alexander, Direct measurement of the local diffusion length grain-boundaries by ebic without a schottky contact, Journal of Physics IV 1 (1991)101–106.

39] D.B. Holt, Local grain boundary property measurements, Solid State Phenomena37–38 (1994) 171–182.

40] D.C. Halls, C. Leach, Processing-induced resistive barriers in ZnO varistor mate-rial, Journal of Materials Science 30 (1995) 2733–2737.

41] J.D. Russell, D.C. Halls, C. Leach, Direct observation of grain boundary Schottkybarrier behaviour in zinc oxide varistor material, Journal of Materials ScienceLetters 14 (1995) 676–678.

42] D.G. Brandon, The structure of high-angle grain boundaries, Acta Metallurgica14 (1966) 1479–1484.

43] Z. Shao, S.M. Haile, A high-performance cathode for the next generation ofsolid-oxide fuel cells, Nature 431 (2004) 170–173.

44] J.S. Yoon, M.Y. Yoon, E.J. Lee, J.-W. Moon, H.J. Hwang, Influence ofCe0.9Gd0.1O2−[delta] particles on microstructure and oxygen permeability ofBa0.5Sr0.5Co0.8Fe0.2O3−[delta] composite membrane, Solid State Ionics 181 (2010)1387–1393.

45] K. Efimov, Q. Xu, A. Feldhoff, Transmission electron microscopy study ofBa0.5Sr0.5Co0.8Fe0.2O3–ı perovskite decomposition at intermediate tempera-tures, Chemistry of Materials 22 (2010) 5866–5875.

46] D. Gao, J. Zhao, W. Zhou, R. Ran, Z. Shao, Influence of high-energy ball milling ofthe starting powder on the sintering; microstructure and oxygen permeabilityof Ba0.5Sr0.5Co0.5Fe0.5O3−[delta] membranes, Journal of Membrane Science 366(2011) 203–211.

47] V.V. Kharton, F.M. Figueiredo, A.V. Kovalevsky, A.P. Viskup, E.N. Naumovich, A.A.Yaremchenko, I.A. Bashmakov, F.M.B. Marques, Processing, microstructure andproperties of LaCoO3−[delta] ceramics, Journal of the European Ceramic Society21 (2001) 2301–2309.

48] V.V. Kharton, A.V. Kovalevsky, A.A. Yaremchenko, F.M. Figueiredo,E.N. Naumovich, A.L. Shaulo, F.M.B. Marques, Surface modification ofLa0.3Sr0.7CoO3−[delta] ceramic membranes, Journal of Membrane Science 195(2002) 277–287.

49] P. Zeng, R. Ran, Z. Chen, H. Gu, Z. Shao, J.C.D. da Costa, S. Liu, Significant effects ofsintering temperature on the performance of La0.6Sr0.4Co0.2Fe0.8O3−[delta] oxy-gen selective membranes, Journal of Membrane Science 302 (2007) 171–179.

50] G. Etchegoyen, T. Chartier, A. Julian, P. Del-Gallo, Microstructure and oxygenpermeability of a La0.6Sr 0.4Fe0.9Ga0.1O3−� membrane containing magnesia asdispersed second phase particles, Journal of Membrane Science 268 (2006)86–95.

51] H. Kusaba, Y. Shibata, K. Sasaki, Y. Teraoka, Surface effect on oxygen permeationthrough dense membrane of mixed-conductive LSCF perovskite-type oxide,Solid State Ionics 177 (2006) 2249–2253.

52] S.P.S. Badwal, Zirconia-based solid electrolytes: microstructure, stability andionic conductivity, Solid State Ionics 52 (1992) 23–32.

53] X. Guo, J. Fleig, J. Maier, Separation of electronic and ionic contributions to thegrain boundary conductivity in acceptor-doped SrTiO[sub 3], Journal of TheElectrochemical Society 148 (2001) J50–J53.

54] J. Maier, Transport in electroceramics: micro- and nano-structural aspects,Journal of the European Ceramic Society 24 (2004) 1251–1257.